Halo Coronal Mass Ejections and Geomagnetic Storms

LETTER Earth Planets Space, 61, 1–3, 2009

Halo coronal mass ejections and geomagnetic storms

Nat Gopalswamy

NASA Goddard Space Flight Center, Greenbelt, Maryland, U.S.A.

(Received June 21, 2008; Revised xxxx xx, 2008; Accepted October 27, 2008; Online published xxxx xx, 2009)

In this letter, I show that the discrepancies in the geoeffectiveness of halo coronal mass ejections (CMEs) reported in the literature arise due to the varied deﬁnitions of halo CMEs used by different authors. In particular, I show that the low geoeffectiveness rate is a direct consequence of including partial halo CMEs. The geoeffectiveness of partial halo CMEs is lower because they are of low speed and likely to make a glancing impact on Earth. Key words: Coronal mass ejections, geomagnetic storms, geoeffectiveness, halo CMEs.

1. Introduction erage speed is ∼1000 km/s compared to ∼470 km/s for or- Coronal mass ejections (CMEs) that appear to surround dinary CMEs), so the V part is expected to be high. When the occulting disk of the observing coronagraphs in sky- CMEs are aimed directly at Earth, the ICMEs are likely to plane projection are known as halo CMEs (Howard et al., arrive at Earth as magnetic clouds (MCs), which are a subset 1982). Halo CMEs are fast and wide on the average and are of ICMEs that have flux rope structure. Since Earth passes associated with flares of greater X-ray importance because through the nose of the MC, the chance of encountering Bs only energetic CMEs expand rapidly to appear above the somewhere within the MC is high (except for unipolar MCs occulting disk early in the event (Gopalswamy et al., 2007). with north-pointing axis—see Yurchyshyn et al., 2001). Extensive observations from the Solar and Heliospheric Ob- When the ICME is shock-driving, the sheath portion lying servatory (SOHO) mission’s Large Angle and Spectromet- between the shock and the MC may also play a significant ric Coronagraphs (LASCO) have shown that full halos con- role in producing geomagnetic storms (see e.g., Gosling stitute ∼3.6% of all CMEs, while CMEs with width ≥120◦ et al., 1990). The sheath may have intervals of north and account for ∼11% (Gopalswamy, 2004). Full halos have south-pointing magnetic fields, in addition to the varying an apparent width (W)of360◦, while partial halos have field orientation in the MC portion. An Earth-directed halo 120◦ ≤ W < 360◦. Halo CMEs are said to be frontsided if CME leads to a situation whereby Earth passes through the the site of eruption (also known as the solar source) can be nose of the shock, where the sheath field is most intense and identified on the visible disk usually identified as the loca- may cause intense magnetic storm if south-pointing. Thus tion of H-alpha flares or filament eruptions. Details on how the ability of a halo CME in producing a geomagnetic storm to identify the solar sources can be found in Gopalswamy depends on the structure of its interplanetary counterpart et al. (2007). Halos with their sources within ±45◦ of the (the ICME event). central meridian are known as disk halos, while those with a Since halos became common place in the SOHO era, central meridian distance (CMD) beyond ±45◦ but not be- there have been several attempts to characterize their geo- yond ±90◦ are known as limb halos. Disk halos are likely effectiveness (see e.g., Zhao and Webb, 2003; Yermolaev to arrive at Earth and cause geomagnetic storms, while limb and Yermolaev, 2003; Kim et al., 2005; Yermolaev et al., halos only impact Earth with their flanks and hence are less 2005; Gopalswamy et al., 2007). Using CMEs from the rise geoeffective (see Gopalswamy et al., 2007). phase of solar cycle 23, St. Cyr et al. (2000) concluded that Since CMEs propagate approximately radially from the ∼75% of the frontside CMEs are geoeffective. In most of Sun (except for a small eastward deflection due to solar the works discussed here, the geoeffectiveness is defined as rotation—see Gosling et al., 1987), disk halos are likely hit the ability of a CME or an interplanetary structure to cause Earth. Of course, the interplanetary counterpart of CMEs geomagnetic storms with intensity (Dst) ≤−50 nT. While (ICMEs) must contain southward magnetic field component most of the intense storms are caused by CMEs, moderate (Bs ) to be geoeffective. It is well known that the intensity storms (−100 nT < Dst < −50 nT) may also be caused of the resulting magnetic storm depends on the magnitude by other structures such as corotating interaction regions. of Bs and the speed V with which the CME impacts Earth’s Zhao and Webb (2003) found an overall geoeffectiveness magnetosphere (see e.g., Gonzalez et al., 1994; Tsurutani rate of ∼64% for frontside halos detected up to the solar and Gonzalez, 1997). Halo CMEs are more energetic (av- maximum in 2000. Yermolaev and Yermolaev (2003) used data from the period 1976–2000 and came up with a lower Copyright c The Society of Geomagnetism and Earth, Planetary and Space Sci- ∼ ences (SGEPSS); The Seismological Society of Japan; The Volcanological Society rate of 40–50%. Michalek et al. (2006) found that 56% of Japan; The Geodetic Society of Japan; The Japanese Society for Planetary Sci- of frontside halos are geoeffective, but they did not use all ences; TERRAPUB. No claim is made to original U.S. Government works. the halos because of limitations in the method of obtaining

1 2 N. GOPALSWAMY: HALO CORONAL MASS EJECTIONS space speeds. Kim et al. (2005) reported that only about all CMEs are full halos, while 11% are full + partial halos 40% of the frontside halos are geoeffective. Yermolaev et (Gopalswamy, 2004). Assuming that these fractions apply al. (2005) compiled published results and noted that that equally well to the front and backside CMEs and recall- the geoeffectiveness rate varied from <40% to >80%. Re- ing that 229 full halos were frontsided, one can estimate cently Gopalswamy et al. (2007) analyzed 378 halo CMEs the frontside wide CMEs as (11%/3.5%) × 229 = 720. covering almost whole of solar cycle 23 and found that Therefore, 720 − 229 = 491 is the likely number of partial ∼71% of frontside halos are geoeffective. Zhao and Webb halos. At the 27% geoeffectiveness rate estimated above, (2003) and Gopalswamy et al. (2007) used the same defini- 134 of the 491 partial halos are expected to be geoeffec- tion of halo CMEs (W = 360◦) and obtained similar geo- tive. Therefore, out of the 720 wide CMEs, 163 full halos effectiveness rates. On the other hand, Kim et al. (2005) and 134 partial halos (total of 297 or 41%) are geoeffec- and Yermolaev and Yermolaev (2003) defined their halos tive. This is virtually the same as the 40% rate obtained by as CMEs with W ≥ 120◦ and obtained the lower geoef- Kim et al. (2005) and confirms that inclusion of partial ha- fectiveness rate. The purpose of this letter is to show that los lowers the overall geoeffectiveness rate. Note that the the discrepancy in the rate of geoeffectiveness can simply geoeffectiveness of halos varies with the phase of the so- be explained by the different definition of halo CMEs used lar cycle (see Gopalswamy et al., 2007, figure 8; Zhao and by these authors. We take Kim et al. (2005) and Gopal- Webb, 2003) and the way multiple halos associated with a swamy et al. (2007) representing the low and high geoef- given storm are treated. We estimate that these effects can fectiveness rates, respectively. Both these works have pub- account for an additional 10% variation. lished their list of events and the study periods have maxi- 2.1 Why are partial halos less geoeffective? mum overlap. They also use data from the same instrument In studying the geoeffectiveness of CMEs as a function (SOHO/LASCO), so the comparison is straightforward. Fi- of source latitude, Gopalswamy et al. (2007) found that nally, both works use the same definition of geoeffective- the strongly geoeffective (Dst ≤−100 nT) and moder- ness: the ability of a CME to produce a geomagnetic storm ately geoeffective (−50 ≥ Dst > −100 nT) CMEs have with Dst ≤−50 nT. average longitudes of W10 and E03, respectively. The non-geoeffective CMEs have an average longitude similar 2. Geoeffectiveness of Partial Halo CMEs to the moderately geoeffective CMEs (E02). Furthermore, Kim et al. (2005) investigated 305 CMEs (1997 to 2003) the fraction of limb halos steadily increases from 17% for that included full (W = 360◦) and partial halos (120◦ < strongly-geoeffective, to 31% for moderately geoeffective, W < 360◦) and found that 121 of them were geoeffective. and 37% for non-geoeffective CMEs. The average speeds On the other hand Gopalswamy et al. (2007) studied 378 also decrease in the same order (see figure 9 of Gopalswamy full halos for the period 1996 to 2005. For making a proper et al., 2007). From these observations, Gopalswamy et comparison, we first separate the full and partial halos in al. (2007) concluded that non-geoeffective CMEs are rel- Kim et al. (2005). We then use the geoeffectiveness of a atively slower, originate predominantly from the eastern subset of full halos from Gopalswamy et al. (2007) corre- hemisphere, and have a greater central meridian distance. sponding to the Kim et al. (2005) study period to estimate Partial halos generally have properties similar to the mod- the geoeffectiveness of partial halo CMEs. erately geoeffective and non-geoeffective halos. Thus par- Among the 378 full halos reported by Gopalswamy et tial halos are less energetic and do not expand enough to al. (2007), 168 occurred during the period 1997 to 2003 fully surround the occulting disk within the LASCO field (the study period of Kim et al., 2005). Many geomagnetic of view. storms during the study period had multiple CME associa- Gopalswamy et al. (2007) also reported that the geoef- tion, so we eliminate 68 such CMEs in estimating the geo- fectiveness rates of limb and disk halos as 60% and 75%, effectiveness rate. Sixty five of the remaining 100 frontside respectively. The geoeffectiveness of partial halos originat- full halos (65%) were geoeffective. Note that this rate is ing closer to the limb is expected to be even lower because close to the one obtained by Zhao and Webb (2003). Ex- they are slower and less likely to impact Earth. Thus, including the full halos from the CMEs in Kim et al. (2005), clusion of partial halos reduces the overall geoeffectiveness we get 205 partial halos (305 minus 100), out of which 56 rate of halo CMEs and hence correctly explains the lower (121 minus 65) were geoeffective. Thus we get a geoeffec- geoeffectiveness rates reported in the literature. tiveness rate of 27% (56 out of 205) for the partial halos alone. The 40% geoeffectiveness rate obtained by Kim et 3. Discussion and Conclusions al. (2005) is thus a consequence of combining highly geoef- One of the important aspects of CME geoeffectiveness fective (65%) full halos and marginally geoeffective (27%) studies is to identify the solar source of CMEs. One might partial halos. argue that the geoeffectiveness may be overestimated when Note that we used the CME data for the period 1997– frontsided CMEs are mistakenly classified as backsided 2003 in the above calculation. To extend this result to CMEs. This can happen only when frontsided halos have the whole study period (1996–2005) considered by Gopal- no disk signature. There may also be cases in which a for- swamy et al. (2007), we need to estimate the faction of tuitous disk activity coincides with a backside CME and partial halos that are geoeffective. Since Gopalswamy et hence a backside CME gets classified as a frontside CME. al. (2007) did not include partial halos in their study, we Sometimes, one observes a full halo, which may be a com- estimate the geoeffectiveness of partial halos by extrapo- bination of multiple CMEs occurring at different position lation. To do this, we make use of the fact that 3.5% of angles. Such occurrences are generally rare and cannot ac- N. GOPALSWAMY: HALO CORONAL MASS EJECTIONS 3 count for the large discrepancies in the reported geoeffec- rates can be readily explained by the different definition of tiveness rate. halo CMEs used by different authors. Partial halos are less To clarify the identification of solar sources of halo energetic and generally originate far from the disk center, so CMEs, let us consider the association between CMEs and most of them behave similar to the nongeoeffective CMEs. soft X-ray flares, one of the obvious indicators of disk activity. Whenever the eruption occurs on the frontside (CMD Acknowledgments. Work supported by NASA’s LWS and TR&T in the range 0 to 90◦), we observe a soft X-ray flare. The programs. halo CME appears asymmetric when the solar source has References a larger CMD, typically beyond 45◦. When the eruption Gonzalez, W. D., J. A. Joselyn, Y. Kamide, H. W. Kroehl, G. Rostoker, B. is behind the limb, but not too far behind, we usually ob- T. Tsurutani, and V. M. Vasyliunas, What is a geomagnetic storm?, J. serve some EUV dimming above the concerned limb, but Geophys. Res., 99, 5771, 1994. no soft X-ray flare is observed because the flare gets oc- Gopalswamy, N., A global picture of CMEs in the inner heliosphere, in The culted by the solar limb. When a flare is partially occulted Sun and the Heliosphere as an Integrated System, edited by G. Poletto and S. T. Suess, chap. 8, p. 201, Kluwer Acad., Boston, 2004. by a limb, the soft X-ray light curve tends to be very grad- Gopalswamy, N., S. Yashiro, Y. Liu, G. Michalek, A. Vourlidas, M. L. ual and we observe the CME above the occulting limb. The Kaiser, and R. A. Howard, Coronal mass ejections and other extreme extreme case is a backside CME whose associated flare is characteristics of the 2003 October–November solar eruptions, J. Geo- completely occulted, and we see no disk activity. In some phys. Res., 110, A09S15, doi:10.1029/2004JA010958, 2005. Gopalswamy, N., S. Yashiro, and S. Akiyama, Geoeffectiveness of of these cases one can see EUV dimming around most part halo coronal mass ejections, J. Geophys. Res., 112, A06112, doi: of the solar disk, indicating a backsided eruption. This kind 10.1029/2006JA012149, 2007. of relationship between the soft X-ray flare and CMEs can Gopalswamy, N., S. Akiyama, S. Yashiro, G. Michalek, and R. P. Lepping, Solar sources and geospace consequences of interplanetary magnetic be easily seen by tracking a large active region (AR) during clouds observed during solar cycle 23, J. Atm. Sol. Terr. Phys., 70, 245, its disk passage and eventual disappearance behind the west doi:10.1016/j.jastp.2007.08.070, 2008. limb (e.g., AR 10486 reported in Gopalswamy et al., 2005). Gosling, J. T., M. F. Thomsen, S. J. Bame, and R. D. Zwickl, The east- When one starts from geomagnetic storms and relate ward deflection of fast coronal mass ejecta in interplanetary space, J. Geophys. Res., 92, 12399, 1987. them to CMEs near the Sun, occasionally it becomes dif- Gosling, J. T., D. J. McComas, J. L. Phillips, and S. J. Bame, Coronal ficult to identify the CME. Zhang et al. (2007) were not mass ejections and large geomagnetic storms, Geophys. Res. Lett., 17, able to identify wide CMEs or their solar sources for ∼10% 901, 1990. of large geomagnetic storms. Zhang et al. (2007) started Howard, R. A., D. J. Michels, N. R. Sheeley, Jr., and M. J. Koomen, The observation of a coronal transient directed at earth, Astrophys. J., 263, with large geomagnetic storms and searched for CMEs and L101, 1982. their solar sources. The solar and geomagnetic events were Kim, R.-S., K.-S. Cho, Y.-J. Moon, Y.-H. Kim, Y. Yi, M. Dryer, S.-C. separated by more than a day to a few days. On the other Bong, and Y.-D. Park, Forecast evaluation of the coronal mass ejec- hand, Gopalswamy et al. (2007) started with halo CMEs tion (CME) geoeffectiveness using halo CMEs from 1997 to 2003, J. Geophys. Res., 110, A11104, doi:10.1029/2005JA011218, 2005. and identified their solar sources. Observations of the halo Michalek, G., N. Gopalswamy, A. Lara, and S. Yashiro, Properties CME and the associated disk signature are nearly simul- and geoeffectiveness of halo CMEs, Space Weather, 4, S10003, taneous. Halo CMEs are more energetic (see figure 4 of doi:10.1029/2005SW000218, 2006. Gopalswamy et al., 2007), so there is usually a prompt (and St. Cyr, O. C. et al., Properties of coronal mass ejections: SOHO LASCO observations from January 1996 to June 1998, J. Geophys. Res., 105, strong) disk signature if the halo is frontsided. Thus, the 18,169, 2000. solar source identification for geomagnetic storms and that Tsurutani, B. T. and W. D. Gonzalez, The Interplanetary Causes of Mag- for halo CMEs do not have the same level of difficulty. Fur- netic Storms: A Review, in Magnetic Storms, edited by B. T. Tsurutani, W. D. Gonzalez, Y. Kamide, and J. K. Arballo, p. 77, AGU Monograph thermore, geomagnetic storms can also be caused by non- 98, AGU, Washington DC, 1997. halo CMEs. If a CME originates close to the disk center Yermolaev, Yu. I. and M. Yu. Yermolaev, Statistical Relationships between and arrives at Earth with a southward magnetic field com- Solar, Interplanetary, and Geomagnetic Disturbances, 1976–2000: 3, ponent, it will cause a geomagnetic storm. It is known Cosmic Res., 41, 539, 2003. Yermolaev, Yu. I., M. Yu. Yermolaev, G. N. Zastenker, L. M. Zelenyi, A. that some magnetic clouds, which are interplanetary CMEs A. Petrukovich, and J.-A. Sauvaud, Statistical studies of geomagnetic (ICMEs) with flux rope structure, are associated with non- storm dependencies on solar and interplanetary events: a review, Planet. halo CMEs. A recent statistical study (Gopalswamy et al., Space Sci., 53(1–3), 189–196, 2005. 2008) finds that only ∼63% of magnetic clouds are associ- Yurchyshyn, V. B., H. Wang, P. R. Goode, and Y. Deng, Orientation of ∼ the Magnetic Fields in Interplanetary Flux Ropes and Solar Filaments, ated with full halos. The fraction increases to 86% when Astrophys. J., 563, 381, 2001. full and partial halos are combined. The remaining 14% of Zhang, J. et al., Solar and interplanetary sources of major geomag- magnetic clouds are associated with non-halo CMEs origi- netic storms (Dst<-100 nT) during 1996–2005, J. Geophys. Res., 112, nating from close to the disk center. Since magnetic clouds A10102, doi:10.1029/2007JA012321, 2007. Zhao, X. P. and D. F. Webb, Source regions and storm effectiveness of constitute the most geoeffective subset of ICMEs, one ex- frontside full halo coronal mass ejections, J. Geophys. Res., 108(A6), pects that some magnetic clouds associated with non-halo 1234, doi:10.1029/2002JA009606, 2003. CMEs are also geoeffective. In conclusion, we confirm that the lower rate of geoef- N. Gopalswamy (e-mail: [email protected]) fectiveness obtained by some authors is due to the inclusion of partial halos. The reported variation in geoeffectiveness